Advancements in single-cell multi-omics technologies are improving our understanding of cellular heterogeneity by providing multiple perspectives on biological systems. Single-cell RNA sequencing (scRNA-seq) is a high-throughput method that measures gene expression at the single-cell level, enabling the exploration of individual transcriptomes and offering insights into cellular diversity and dynamic biological processes. In contrast, CITE-seq (Cellular Indexing of Transcriptomes and Epitopes by sequencing) is a multi-omics approach that links cell states to functions by simultaneously profiling RNA and surface proteins from the same cell.

Mosaic data integration is used to merge multi-modal single-cell data (e.g., scRNA-seq and protein data in CITE-seq), creating a unified representation across different data types. Since various technologies capture distinct cellular aspects-such as gene expression, protein abundance, and chromatin accessibility-mosaic integration align them into a common latent space while maintaining biological relationships and mitigating batch effects introduced by differing experimental protocols. Integrating RNA and protein data enhances our understanding of post-transcriptional regulation, particularly in cases where RNA and protein expression levels do not correlate.

The goal of this assignment is to develop a computational method that learns latent space representation for gene expression and protein datasets and performs clustering on the latent dimensions. To learn more details on the problem, read the following paper.